Background
The identification of extracellular vesicle (EV)-associated biomarkers is crucially dependent on methods that allow the isolation of EVs without contaminating plasma proteins and lipoproteins, yield a sufficient quantity of EVs for downstream molecular analysis, and are reproducible, efficient and easy to perform.
Isolation of EVs from plasma is prone to encounter several obstacles related to high plasma viscosity, high lipid and protein content and the presence of platelets. Frequently, the sample size is a limiting factor too. Currently, standardized protocols for isolation of EVs from plasma are missing.
The most popular method for EV isolation, differential ultracentrifugation, suffers from several limitations (impurities, EV aggregation, decreased integrity of EVs) and impracticalities (long turnaround time, specialized equipment) [
1,
2]. The growing interest in blood-based EV-associated biomarkers in recent years has led to the development of novel EV isolation methods that could be well suited for clinical research. Despite the growing number of these methods, the field suffers from a lack of standardization and often, new techniques are used without detailed comparative analysis.
Size exclusion chromatography (SEC) has been shown to perform well in separating EVs from contaminating plasma proteins and high density lipoproteins (HDL) [
1,
3,
4] and has been successfully used for small scale analysis of EVs from clinical samples [
4,
5]. Despite becoming the method of choice for isolation of relatively uncontaminated EVs from plasma, SEC has several technical and practical limitations. SEC only permits efficient isolation of EVs larger than the pore size of the matrix used (i.e. 70 nm for CL-2B Sepharose). Although free of HDLs, EV-rich fractions can still contain a small amount of other lipoproteins such as chylomicrons (100–600 nm) and very low density lipoproteins [VLDL (30–80 nm)] [
4,
6,
7]. Although the sample processing time is short (20 min) compared to differential ultracentrifugation, SEC still requires considerable hands-on time for the preparation of the isolation column (if homemade), washing and (re)equilibration. In addition, manual collection of fractions may introduce operator-based variability, especially affecting the purity of the fractions. Other limitations of SEC include the relatively low vesicle yield and the dilution of a purified sample which requires an additional concentrating step that may result in a yield drop [
3].
Recently, a new membrane affinity spin column method for the isolation of highly pure EVs from biofluids was released (exoEasy kit from Qiagen) [
8] that appears to have the potential to overcome many of the limitations of SEC. In addition, EV isolation can easily be coupled with RNA extraction directly from the EVs bound to the column membrane (exoRNeasy Serum/Plasma kit), hence offering a simplified workflow for downstream analysis of the RNA content of EVs, which could facilitate clinical research into novel EV-associated RNA biomarkers.
Here we have compared the large sample volume version of the exoEasy kit, the exoEasy Maxi kit, with the SEC using qEV 10 ml columns (Izon Science) for isolation of EVs from 2 ml of human plasma. We have focused our analysis particularly on ELVs, the clinically most relevant EV subset. We have evaluated the yield, size distribution and purity of isolated EVs as well as their RNA content including size range and yield.
Our data show that the exoEasy kit isolates a heterogeneous mixture of particles with a larger median diameter, broader size range and a higher yield than the SEC qEV column. The exclusive presence of small RNA in the particles and the total RNA yield were comparable between the kits. Despite being less prone to low density lipoprotein (LDL) contamination than the SEC qEV column, the low particle-protein ratio, significant amount of albumin, very low levels of exosome-associated proteins and propensity to triglyceride-rich lipoprotein contamination suggest isolation of mainly non- ELVs and co-isolation of plasma proteins and certain lipoproteins by the exoEasy kit.
Methods
Sample collection and preparation of platelet-free plasma (PFP)
Two groups of six healthy anonymous donors and three lymphoma patients were included in the study. Blood was drawn from non-fasting donors into K
2 EDTA tubes (BD Vacutainer) and processed within 2 h of blood draw. It was centrifuged at 2500×
g for 15 min at 20 °C. Cell-free, platelet poor plasma was collected and subjected to the second centrifugation under the same conditions. The supernatant was finally centrifuged at 13,000×
g for 5 min and the resulting PFP was filtered through a 0.22 µm filter, aliquoted and stored at − 80 °C [
9]. Prior to use, plasma was quickly thawed in a water bath at 37 °C and clarified by centrifugation at 10,000×
g for 20 min to remove apoptotic bodies and microvesicles. The first set of six plasma samples from healthy donors was processed as described above. For the second set of plasma samples, collected from additional six healthy donors and three lymphoma patients and used primarily for determination of total lipid levels of plasma EVs, plasma ultrafiltration and high-speed centrifugation were omitted.
EV isolation from plasma using SEC qEV columns
After rinsing the columns with PBS, 2 ml of PFP were applied on top of a qEV column (Izon Science) and 0.5 ml fractions were collected. Four EV-rich fractions (7–10) were pooled and either analyzed directly (see below) or concentrated using an Amicon Ultra-4 10 kDa centrifugal filter device (Merck Millipore).
EV isolation from plasma using exoEasy kit (Qiagen)
This was performed from 2 ml of PFP according to the manufacturer’s protocol with modifications described in Enderle et al. [
8]. EV eluates were either analyzed directly (see below) or concentrated as described above. EV eluates from both the SEC qEV columns and the exoEasy kit were aliquoted in low protein binding tubes and single use aliquots were stored at − 80 °C.
Total protein quantification and Western blot analysis
Proteins were extracted from concentrated pooled EV-rich SEC qEV fractions or exoEasy kit eluates using a lysis buffer (1% NP40, 1 mM EDTA) with a protease inhibitor cocktail (Roche). Samples were vortexed and lysed on ice for 15 min. The total protein content of EVs was measured by Pierce BCA Protein Assay Kit (Thermo Scientific). To assess the purity of EV preparations, a ratio of number of particles to micrograms of protein was calculated [
10]. The presence of EV-enriched proteins as well as the absence of endoplasmatic reticulum (ER) markers and plasma proteins was determined by Western blotting in 10 µg of lysates using the following antibodies: syntenin-1 (gift from P. Zimmermann), Tsg101 (612697, BD Biosciences), CD63 (556019, BD Biosciences), CD81 (sc-166028, Santa Cruz), calnexin (sc-2679, Santa Cruz) and albumin (4929, Cell signaling).
Total lipid quantification
Lipid content of EV preparations was determined by sulpho-vanilin assay using the Lipid Quantification Kit (Cell Biolabs, San Diego, USA) under conditions optimized for EV analysis [
6].
Determination of plasma lipoproteins
Levels of plasma triglycerides, total cholesterol, HDL, LDL, non-HDL cholesterol, apoA1 and apoB were determined in a clinical diagnostic laboratory of University Hospitals Leuven on a Roche Cobas 8000 chemistry analyzer.
Transmission electron microscopy (TEM); negative staining
Aliquots from pooled EV-rich SEC fractions or exoEasy kit eluates were deposited onto formvar-coated 400 mesh copper grids for 7 min at room temperature and thereafter stained with 1.75% uranyl acetate. The grids were observed using a transmission electron microscope JEM 1400 (Jeol Ltd.).
Nanoparticle Tracking Analysis (NTA)
This was performed using a NanoSight LM10 instrument (Malvern Instruments Ltd.). Aliquots from pooled EV-rich SEC fractions or exoEasy kit eluates were diluted in filtered PBS. Six videos of 30 s were captured for each sample.
RNA isolation from EVs
Total RNA was isolated from concentrated pooled EV-rich SEC fractions or directly from exoEasy kit columns following the protocol of the exoRNeasy Serum/Plasma Kit (Qiagen). The protocol was optimized to increase the yield by addition of glycogen (5 µg/ml), double extraction of the aqueous phase [
11] and double elution of RNA. RNA yield and size range were analyzed on an Agilent 2100 Bioanalyzer using the RNA 6000 Pico Kit and the Small RNA Kit (Agilent Technologies).
Statistical analysis
Data were analyzed with the GraphPad Prism software using Wilcoxon matched-pairs signed rank test with p values * < 0.05 and ** < 0.01. Data are presented as mean ± SEM.
Discussion
The growing interest in molecules carried by EVs as potential circulating biomarkers has spurred a lot of research and development into the methods for isolation of plasma EVs, especially ELVs. The isolation of a high yield of high-purity ELVs from plasma is technically challenging especially due to its high density and viscosity and the complex composition [different types of vesicles, proteins, ribonucleoproteins (RNP), lipoproteins]. The methods and protocols to achieve this goal are thus under constant development.
Recently several novel methods have emerged in the form of a kit that enable an easy and fast isolation of EVs from plasma. Among EV isolation methods potentially suitable for an EV biomarker discovery we identified exoEasy kit (Qiagen), based on its primary evaluation by Enderle et al. [
8], to be a potentially promising, faster and easier-to-use alternative to SEC, the current method of choice. EV isolation by the membrane affinity spin column of exoEasy kit is based on a universal biochemical feature specific to EVs. Other particles, such as plasma proteins, lipoproteins or RNP complexes that often contaminate EVs prepared by classical methods, should not be co-isolated. The broader size range expected for EVs isolated by the exoEasy kit compared to the lower particle size limit of 70 nm for EVs isolated by SEC qEV columns could improve the detection of low abundance biomarkers and/or enable detection of small size EV-based biomarkers. Different column sizes allow the isolation of EVs from plasma volumes ranging from 0.1 to 4 ml. The option to extract RNA directly from column-bound EVs (exoRNeasy Serum/Plasma kit) is another feature that makes this kit attractive for potential clinical use. The time required for the purification of EVs using the exoEasy kit (25 min) is largely the same as required for the SEC qEV columns without the additional hands-on time for qEV column washing and re-equilibration. ExoEasy kit columns cannot be re-used, but enable processing up to 4 ml of plasma. Although plasma volumes processed by a 10 ml SEC qEV column vary in literature, only 0.5 ml is recommended by the manufacturer. The SEC qEV columns however, can be re-used up to five times thus enabling processing of up to 2.5 ml of plasma.
In this report we provide a comprehensive assessment of the exoEasy kit for the isolation of EVs, particularly ELVs, from human plasma, focusing on its comparison to the currently very popular SEC qEV columns. We used multiple complementary techniques for this purpose [
19,
22].
The results of TEM and NTA analyses indicate that exoEasy kit isolation results in a highly concentrated sample containing a heterogeneous mix of particles that differ greatly in size and shape. Both analysis methods show that the majority of particles isolated by affinity spin columns were significantly larger than particles isolated by the SEC qEV columns, with a diameter exceeding the typical size range of ELVs (30–150 nm) [
23]. The difference in the EV sizes measured by NTA and TEM were expected given that particles are analyzed in a hydrated and desiccated state, respectively. The hydrodynamic diameter of ELVs was reported to be up to three times larger than the geometric one [
24]. In line with our data, others also have detected the presence of larger particles in eluates from exoEasy kit when using TEM and light scattering methods as a readout [
25]. These larger particles were attributed to the kit elution buffer. Our NTA analysis of elution buffer however did not confirm these findings.
Previous studies have highlighted the differences in EV diameter between patients with classical Hodgkin lymphoma and healthy donors [
5,
26] with patients displaying particles with a smaller diameter. The propensity of the exoEasy kit to isolate larger vesicles might lead to a suboptimal isolation of the population of small EVs from the patients’ plasma and might thus affect the subsequent biomarker analysis. Similarly, the SEC qEV column cut-off is 70 nm and thus it is also likely to inefficiently isolate smaller EVs. While here we focused primarily on evaluation of the exoEasy kit’s the capacity to isolate ELVs from plasma, the isolation of a broad range of vesicles by the exoEasy kit could be useful in certain applications.
A high particle concentration was detected by both NTA and TEM in exoEasy eluates, however it was not reflected by the number of ELVs observed by TEM. Instead, the presence of proteins and large size particles was readily detected upon sample dilution. Given that NTA does not differentiate between vesicles and non-vesicular particles/aggregates, the higher particle concentrations detected in EV preparations from exoEasy kit could reflect the presence of RNP complexes, protein aggregates or lipoproteins.
Lipoproteins are a major subcellular particle subset of plasma [
27], thus it is not surprising that they are frequently found in EV preparations from plasma. The presence of contaminating lipoproteins in EVs could potentially affect certain downstream molecular analyses For example HDLs were identified as circulating miRNAs carriers [
28] Currently, no stand-alone EV isolation method results in a complete removal of plasma lipoproteins. SEC efficiently removes contaminating proteins (< 1%) and HDLs (< 5%) from plasma EVs [
3‐
5] however recent reports have shown that lipoproteins, especially LDLs (~ 25 nm, but also aggregates 100–600 nm) might co-isolate in exosomal fractions retrieved from the SEC qEV columns [
3,
6,
27]. Indeed, we have observed small spherical particles in our SEC qEV preparations with a diameter corresponding to that of LDL. In addition, we obtained evidence for co-isolation of LDLs and triglyceride-rich lipoproteins (chylomicrons, VLDL and their remnants) in SEC EV preparations based on the total lipid analysis of EVs from plasma of subjects with distorted plasma lipoprotein profiles. It is assumed that lipoproteins do not co-isolate with EVs during exoEasy kit-based EV isolation [
8], however this was not previously excluded. Here we demonstrate that exoEasy kit is more effective in removing contaminating plasma LDLs from EV preparations than the SEC qEV columns, while still showing propensity to contamination with triglyceride-rich lipoproteins. Since our other observations including the presence of vesicles in the size range of VLDLs, a high protein content and a low particle-protein ratio could serve as additional indirect evidence for the presence of co-isolated lipoproteins, further evaluation of co-isolation of other lipoprotein classes is warranted.
In a previously published evaluation of the exoEasy kit only a single vesicle marker was used to detect the presence of EVs [
8] and the purity of the EV preparations was not assessed. In addition, the manufacturer’s recommendation to concentrate the isolated EVs using a protein filter with ≤ 100 kDa pore size may mask the presence of contaminating plasma proteins. Focusing our protein analysis on a recently updated set of proteins characteristic of ELVs we confirmed the high enrichment of these markers, including syntenin-1, TSG101, and CD81 in samples from SEC qEV columns. In a stark contrast, however, these proteins were either undetectable or present only at very low levels in exoEasy kit samples. The presence of albumin indicates an insufficient purity of the EV preparation, as albumin has been shown not to be part of the EV proteome [
29]. Albumin co-isolated with EVs from exoEasy kit samples at much higher levels compared to SEC and together with the low particle to protein ratio was another indicator of inferior purity of exoEasy kit EVs. We and others have detected low levels of albumin in vesicle-rich fractions eluted from SEC qEV columns [
14]. We should point out that the pool of fractions 7–10 was used in this analysis. Using a pool limited to fractions 7–8/9 results in lower levels of contaminating albumin (manufacturer’s instructions). qEV columns contain the CL-2B Sepharose. Recently other matrices (Sepharose CL-4B or Sephacryl S-400) were shown to perform better in the separation of EVs from albumin [
14].
Authors’ contributions
RS designed the study. RS, LG, JW, KB, PV and DD performed the experiments, collected and analyzed the data RS, GA and RS wrote the manuscript. All authors read and approved the final manuscript.